Market-driven is Pragmatic Institute's term for making product decisions from evidence collected outside the organisation. The alternative it is defined against is not malice or laziness, it is the ordinary situation in which the people who decide what gets built are the people furthest from anyone using it.
Where product decisions usually come from
The default sources are all internal and all plausible.
- An executive has a conviction.
- A large customer escalates.
- A competitor announces something.
- Sales loses a deal and reports a missing feature.
- Engineering proposes a rewrite.
Each of these is a real signal and none of them is a representative sample. The loudest of them is usually the one attached to the most senior person in the room rather than the one describing the largest problem.
The framework's answer is not to ignore them but to route them through the same test. What market problem is this, who has it, how urgent is it, how many people have it, and would they pay to have it solved.
The noisy twenty and the quiet eighty
Foundations uses a specific framing for the sampling problem. A small proportion of customers generate most of the inbound signal, through support tickets, escalations, user groups and account reviews. They are not typical. They are the ones with the time and the relationship to complain.
The implication is uncomfortable. A backlog built entirely from inbound requests is a backlog built for the minority who shout, and the majority who quietly stop renewing never appear in it. Structured discovery exists to reach the people who do not volunteer, which means going to them rather than waiting.
The habits that make it real
Regular contact with the market on a schedule. A number of interviews per month, held to the way a release date is held. It is the first thing dropped under pressure and the first thing that should not be.
Interviews with people who did not buy. Buyers tell you why your product worked. Non-buyers tell you what the market considered and rejected, which is the more expensive information to lack.
Problems recorded separately from solutions. A market problems table that holds the problem, who has it, evidence of urgency and evidence of pervasiveness keeps the record usable after the person who ran the interview leaves.
Decisions traced back to evidence. Every item in a roadmap should be answerable with the question of which market problem it addresses. Items that cannot answer it are not necessarily wrong, but they are opinions and should be labelled as such.
What it changes in practice
The visible difference is in how disagreements end. In an opinion-driven organisation they end when the most senior person stops talking. In a market-driven one they end when someone produces evidence, or when everyone agrees that the evidence needed does not exist yet and names what it would take to get it.